ECL 224: BART "Dead Train" Detection

Engineering Case Library

ECL 224

BART "DEAD TRAIN" DETECTION

Four students learned about one
of the many difficulties which
beset the start-up of the San
Francisco Bay Area Rapid Transit
system. They relate what they
learned from interviews and add
their comments and evaluation.

The contents of this case and the statements, views, and conclusions are
those of the University of Santa Clara personnel and are not attributable
to the San Francisco BART District.

ECL 224

Daniel Y. N. Chu
John A. Paloroo
James F. Rollings

Prepared under the supervision of R. K. Pcfley, Mechanical linpinccring Dept.

— - SCHOOL OF ENGINEERING

ENGINEERING AND APPLIED SCIENCE RESEARCH
UNIVERSITY OF SANTA CLARA

SANTA CLARA, CALIFORNIA 95053 Tel. No. (408) 984-4325

THE CONTENTS OF THIS REPORT AND THE STATEMENTS, VIEWS AND CONCLUSIONS EXPRESSED
HEREIN ARE THOSE OF THE UNIVERSITY OF SANTA CLARA PERSONNEL AND ARE NOT ATTRIBUIABLE
TO THE SAN FRANCISCO BAY AREA RAPID TRANSIT DISTRICT.

ECL 224

As with any transportation system whether large or small, complex or
simple, safety is an essential and uncompromising factor in the design and
implementation. Yet safety is one of the most difficult requirements to
define for any system. The concern of this paper is a safety-related
problem of the San Francisco Bay Area Rapid Transit (BART). Specifically,
the problem is one of train detection which is vital to the automatic train
protection.

From its beginning, BART planned to operate trains by automatic means,
not as a labor saving method but as an operating necessity: the control of
high-speed trains (up to 80 mph) on close headways was beyond human ability
to do so safely and consistently. Such a scheme must be able to detect train
presence and position in relation to leading and following trains in order to
maintain proper speed regulation and sufficient train separation. Detection
is vital to passenger and train safety.

The train detection problem is not a hard problem to describe as an
isolated and technical deficiency. But a discussion on the implications,
the possible solutions, and the development of the problem is much more
involved.

In order to fully understand the train detection problem it is necessary
to explain how BART was conceived, how the automatic train control was
developed, what tests and safeguards were set up and what these tests were
lacking. It is necessary to go back to the beginning. And, for BART, the
beginnings are in the late 1940 ! s and early 195G ! s.

In the postwar era, the San Francisco Bay Area experienced an enormous
growth in population. People came to the cities of the Bay Area looking for
jobs and a place to live. What developed is the urban and suburban sprawl
which makes up the greater San Francisco Bay Area. With that sprawl came the

2 ECL 224

rapid growth of automobile commuter traffic. l;very working day, tlvp cities
of San Francisco, Oakland and Berkeley were the sites of a mass convergence
in the mornings and a mass exodus in the evenings. The commuters came from
the outlying areas of the eastern counties of Contra Costa and Alameda, from
Marin County in the north and from San Mateo County in the south. Having no
adequate alternatives, people commuted by private automobile, resulting in
massive congestion on the freeways and in the cities.

The geographic features of the area, such as the Bay, limited access
between major city centers, causing the bridges to become traffic bottlenecks.
In 1947 the Navy Civil Engineering Corps and the Army Corps of Engineers
recommended the construction of a transit tube beneath the Bay to alleviate
the congestion. Studies predicted even more population and traffic growth.
The only foreseeable solution was some type of large-scale, high-speed transit
system. It would have to be something that would get the daily commuter out
of his car*

In June 1957 the California State Legislature authorized the creation
of the Bay Area Rapid Transit District (BART). The district originally
included the five counties of San Francisco, Contra Costa, Alameda, San Mateo
and Marin, (In 1962 Marin and San Mateo counties withdrew from the district
under economic and public pressures.)

In May of 1959 the BART district retained three prominent engineering
firms (Parsons, Brinckerhof f , Quade and Douglas; Tudor Engineering Company;
and the Bechtel Corporation) to act as the general engineering consultants
and contractors for a comprehensive rail mass rapid transit system. They were
to operate as a single unit under the name of Parsons-Brinckerhof f-Tudor-
Bechtel (PBTB) , By 1962 PBTB had submitted its general proposal for BART.
It was an ambitious plan that would take ten years to complete at an

3 ECL224

estimated cost of one billion dollars* This ambit iousncss was expected,
considering the scope of the problems it hoped to solve.

Thus, BART hod its start as a saving solution to the Bay Area's traffic
problems. Yet, before a single passenger revenue dollar was to be collected,
BART was to face crisis after crisis of economic, political and technical
problems. The financing of the one billion dollars presented a tremendous
economic problem coupled with political and public opposition to the whole
scheme.** The trans-bay tube and the seismic activity and the soil conditions
of the area presented challenging problems to the structural and civil
engineers. Arid, the concept of complete automatic train control was not
without its problems. One such problem is the 'Head trairi' detection problem.

*In the three counties, the $762 million bond issue to be used to finance
BART passed by a narrow margin of 1.2% over the required 60%.

**This is the estimated cost of the three county system.

4 ECL 224

The Test Track Program
The concept of automatic train control could not precisely be called
novel. It could be said to have been unprecedented on such a large-scale.
It should be noted that BART was the first rapid transit system to be
constructed in North America in many years. Because of the lack of a
market for new rapid transit equipment, manufacturers were not spending
much on research and development of systems or equipment for rapid transit.
Rapid transit technology was basically that of the 1930 f s, whereas in
almost every other field of technology and engineering work there had been
tremendous advancements. It was BART's objective to bring this improved
technology to bear on the BART system. This was the philosophy behind the
Test Track Program (1964-66) and behind PBTB's solicitation to more than 15
manufacturers to propose methods of fulfilling the general functional require-
ment for automatic train control. These requirements were fairly broad to
allow the manufacturer to respond on a broad conceptual basis. More specific
requirements would be drawn up after the test program. In fact, one of the
objectives of the Test Track Program was to form a demonstrative basis from
which PBTB could prepare the specific functional requirements for the ultimate
system.

Seven firms replied to PBTB's solicitation. Of these, four systems were
found sufficiently complete to warrant their demonstration in the Test Track
Program. The four firms that participated in the program were General Electric
Company (GF.) » General Railway Signal Company (GRS) , IVcstinghouse Hlectric
Corporation (W1ZLC0) , and Westinghouse Airbrake Company (WABCO).

As expected, since PBTB specified only functional requirements, the four
companies presented diverse techniques with which to fill the general require-
ments. A brief overview of the competing control systems is shown in

5

ECL 224

Table 1, which has been taken from an article printed in Hlcctronics „

Magazine, July 26, 1965. At this point, let it suffice to say that the four

systems "successfully met the intents of the General Functional Requirements

for these demonstrations". Yet, no one system was singularly outstanding and

while there were successful demonstrations, !, no single system fulfilled in

entirety all the requirements for the RART automatic train control system".

The Test Track Program consisted of a series of tests which can be broken

down into three categories:

Part I: "Normal circulation" tests in which the manufacturer
demonstrated ability to operate the interlockings , to stop at
stations, to control doors, reverse automatically, and to observe
speed limits as required. If a contractor failed any test, he was
allowed to check out the system, correct the problem and redemonstrate.
Upon successful completion of all of these tests, Part II was started.

Part II: "Disrupt" tests. In these tests, willful safety violations
were made, such as disconnecting all or part of the contractor's
station-stop controls to test the ability to operate safely under
these conditions. The presence of a train was simulated to check the
enforcement of train separation. If these tests were completed
satisfactorily, the "chase" took place. In the "chase" the car being
tested was operated automatically behind a slow-moving car operated
manually (except in the case of WELCO's demonstration in which the
lead car operated automatically- -manual operation on the main line
was against the inherent properties of the system and would have shut
it down). The lead car changed its speed and stopped in accordance
with commands with the test coordinator to check for the safe response
of the test car.

Part TIT: "Special" tests in which the contractor demonstrated special
features of his system.

The test program served its purpose fairly well, though in hindsight
may be viewed as being deficient in an aspect that might have led to the early

ECL 224

CO
Q

s

2£

s

H

x;
o
u

z

►H

a.

8

a.
o

uj

a:

a)

u
a>

c:

O rH

U

Q

r-H

J-

u

•h

cd <x>

rH

r— 1

O

CD

0)

. — i

*->

G *->

+->

o

•H

O

4->

CO

+-)

C

#-J

cd

U o o

cd

li

U

rd

c

to

Cl

4J

-r-H

CL

■rH

o

to

u

V)

CU

CD

o

fc o

;i

to

p«

£

00

•H

i-H

cd O -t

rj

T~H

<Q

o

LO

O LO

o

u

r— <

cd

o

>.

CO

•H

u

CO

•rH

ce

a

4-J

CO

•fH

u

/ V

LO

o

C U

+-»

r-«

o

c

*-»

CJ

u

+-J

U

•H

o

c

•H

a

o — i <£

•h nj Du

•H

cd

u

o

o

rj

•H

U

u

o

C-l

to

E

>

• H

CD

"J

TJ

E:

>

• H

cd

1

C

TJ

4-> ^ \

to

•H

•o

+->

CL

•H

o

>h O

to

■ H

•M

Cl.

CD

ct3 +-> tU

<D

o

ct3

cd

c

to

rH

*-> U

C

CD

cd

o

cd

Cj

cd

C

cd

U

c

<->

o

^»

a.

C CL

cd

u

r-H

c

3

4->

Q

rH

LO CD * — '

r-l

<D

CO

rH

rt

^

Re

CD

to

0)

S 1

u

a:

to

u

H

c£

H

U

•H

rH

O CO

H

fj p.

3 3

C

o

E u

•H

O 'H

K

U CL

•H

W) rH

O «1

CTJ U)

U

C «H

< to

•H

rH

r-H

o

4->

c

o

u

O

<u

a

to

o
o

rH

c ^:
W) u

to H

60

c

a:

r^

0) tO

^ CL

E u

O »h

U CL

Wj rH

O al

C *H

< CO

al
U

c
o
a

o

X
■H
LL

to

*H

u

c:

U

cr

o

r-l

to

Uh

rH

O

•H

W)

•H

<

to

1-4

o to

+-> CL 1-4

d d fl)

rH ^ 4->

CD U CD
C *H E
CD CL O

t>0 rH
rH CD
O Cd r-H

•H C CD

TJ W) -J

< to <

a)

r-«

c

CD
U

<D
X

LL

0)

u
<d

D c:

cr o en
o ^

r-l to

CL
O

o

<4-l rH CD r-H

CTJ rH

O C cd CD

•h bo D rH

D to to

<

1-4

CD to
*-> CL

CL^i

G u

O -H

a cl

U rH

o cd
c

cd bo

<: t/j

CD

u

cd

1-4

c

<D
U
CD
Cl

CD
X
•H
LL

to

. — (

►H

u

rt

c:

rH

CD

c

cr

o

CD

rH

to

Uh

rH

cd

O

c:

-H

bO

u

•H

H

to

<

0Q
<

o

u

•H

>,

rH

c

. «r-»

cd

U

CL

CD

e

. rH

o

UJ

u

rH

cd

rH

CD

C

CD

cd o

cj cd

u c

<L> b0
C •H
CD OO

CL

1-4

CD O
to C_J

o u

JC -H
bO rH

•H U

4-> CD

tO rH

CD LU

O

CD O

to

3 CD

O r^

-C cd

b0 rH
C CO
•H

4-> rH

CO 'rH

CD <

7 ECL 224

discovery of the "dead train" detection problem. Yet, the primary pur|u>se of
the test track program was more of a feasibility study rather than a system
selection process. In fact, it was not necessary for a manufacturer to
participate in this program to be eligible to compete in the bidding for
supplying and installing equipment for the BART system. The reason for this
was that not all companies interested in making bids would have been able to or
would have been permitted to participate in the test program due to limited
funds. (The test program was funded by a Housing and Home Finance Agency grant.)

From the results of the test program the Standard Specifications for the
San Francisco Bay Area Rapid Transit were drawn up. Part J of the Standard
Specifications concerns the automatic train control (ATC). Three very
important articles of the Standard Specifications concerning train detection
are quoted below:

Article J. 6. 3 Train detection shall be employed to detect the presence

of trains for train separation and route interlocking
functions.'

Article J, 6. 3.1 Train detection shall detect the presence of trains

throughout the entire revenue system. Trains shall be
detected continuously. The maximum length of a train
detection zone shall be 5,000 feet.

Article J. 6, 3.2 If train detection equipment is incapable of detecting

the presence because of momentary loss of signal or
other conditions, the zone or zones which the train
detection equipment is supervising shall be indicated
as occupied.

Another article of the Standard Specifications stated that a detection
system must be provided for protection against foreign vehicles or objects on
the right of way.

As can be seen, these Standard Specifications are in the form of
functional or performance specifications. It was left up to the contractor

8 ECL224

to translate the performance specifications into hardware specifications.
This was in keeping with the philosophy of the Test Track Program to allow
diverse techniques and construction costs.

The BART District called for and received on February 28, I9G7,
competitive bids for the main line ATC and communications system. On March 16,
1967, PBTR recommended to the District that the ATC and communications contract
be awarded to the low bidder, Westinghouse Electric Corporation. The low
bid was $26,199,959.32.* The WELCO proposal that was approved by PBTB was
quite different from the system WELCO demonstrated in the Test Track Program
in 1964-1966. As stated before, participation in the test program was not
a prerequisite for bid selection. One might say that PBTB's approval was
M on paper", taking into consideration the merits of the proposal itself and
WELCO 1 s experience and test results.

It had been suggested by PBTB that a successful bidder be required to
install his system on the test track and laboratory cars to demonstrate the
successful operation of his system prior to the installation on the remainder
of the system. Limited testing on the main line with laboratory cars was
carried out.

The Westinghouse System
As mentioned previously, the system that BART bought was significantly
different from the system WELCO demonstrated in the Test Track Program. Lower
construction and maintenance costs were cited as reasons for the change. The
following is a synopsis of that new system—how it should have worked and why it
doesn't work:

The system employs the method of dividing the track into train protection

blocks. Each block varies in length to a maximum of 1,200 feet. Each block is

*This was approximately $6.4 million dollars below PBTB's estimate. Yet,
by July 17, 1973, change orders and escalation costs increased the cost to
$34,943,176.53. The next lowest bidder was WABCO at approximately s ^29.6 million.

9 ECL224
a track circuit loop formed by the running rails and heavy rail to rail
bonding. A sketch of the track circuit is shown in Figure 1. At the ends of
each circuit a transmitter, a receiver and a multiplexer were housed. Simply
and basically, the track circuits perform two functions: (1) speed limit
control and (2) train detection necessary to enforce power train separation.
Each transmitter sends one of three audio frequencies (to distinguish adjacent
track circuits) each with a digital modulation representing the speed command.
The train equipment picks up the speed command from the rails through an
inductive coil. The receiver at the other end of the block also picks up the
signal, if it is present through inductive coupling. The transmitted signal
is compared with the received signal at the local equipment room. If the
signal comparisons agree, the circuit is assumed to be unoccupied. If a
train is within the block, the steel wheels and axles will shunt the signal and
the signal comparisons will not agree. The block is then considered occupied.

The train protection is provided by the local wayside equipment and not
by the central computer. The maximum speed limit for each block is stored in
the local equipment and it is also from this equipment that speed commands,
enforcing proper separation, originate. The central computer performs the
duties of scheduling and routing verification.

Tests prior to and after the opening of BART for revenue service in 1972
showed that the system worked fine if the train was detected. But, in the
worst case, trains could not be detected. This is the problem of "dead train"
detection. The worst case occurs when there is no power being supplied to the
train and there is contamination on the rails (rust). Momentary loss of
detection also occurs even with powered trains.

The wheel-to-rail, when there is contamination, take on the characteristics
of a diode (see Figure 2). When there is power supplied to the train, the train-
to-track potential exceeds the breakdown voltage of the rail contamination.

10

o

o S

o -J

cr cr

UJ UJ

X X

UJ UJ

_J _J

Q- Q.

3 3

ECL 224

q:

UJ

a:

X

UJ

UJ

>

-J

UJ

CL

h-

-J

CO
m

o

N
O

q:
x
o

>-
co

EXER

ITTER

CL

CO

Li

z

<

a:

<

CO

>-
a

u

a

Ul

cr

x

o

X

CO

o
a:
o

—i
o

o

CO
Q

a:
o

o
q:
h-

O

o

Q
<

CO

GO
O

NJ
Z

o
a:
x
a

>

CO

o

s

CO

Q
UJ

CL
O
O

CO

>-
<
_J

Ul

a:

ui
<

(0

I
o
I—
UJ

CO

CO
2

o

h-
o

UJ
2
2

o
o

Q
<

3
O

cr
o

<

I-

Ui

cr

3
O

u.

£ oz

< OS

o

Ul

11

ECL 224

CURRENT

VOLTAGE

FIGURE 2 • RAIL TO WHEEL IMPEDANCE

CHARACTERISTICS

12 ECL 224

This breakdown of the contamination results in a low impedance and the control
signal is shunted by the train. On the other hand, when there is no power to
the train that is a "dead train", the contamination causes a high impedance.
This may keep the control signal from being shunted which in turn causes this
particular block to be labeled incorrectly as "unoccupied". As can be seen,
this presents a potentially dangerous situation.

This "dead train" detection problem was discovered in 1971-72 just prior
to the opening of the revenue service. In 1971-72 the Public Utilities
Commission of the State of California (PUC) , which has safety jurisdiction
over BART, insisted that BART prove that the system could detect "dead trains".
Tests were performed and BART failed.

The question now is: how could such a problem in a publicly funded
project be overlooked by so many over such 'a long period of time (almost seven
years)? The involved parties were Westinghouse Lilectric Corporation, Parsons-
Brinckerhoff-Tudor-Bechtel, and the Public Utilities Commission. The result is
a multimillion dollar law suit filed by the BAPT District against PBTB, WCLCO,
and others, not only for this particular defect but for others as well.

To begin with, in the Test Track Program, "dead train" detection was
never performed. It was not a test consideration, even when a malfunction of
WABCO's system, which is employed through the rail control signals, occurred
due to rail contamination. It is our speculation that this evidence was
ignored, "while WABCO was performing the disrupt portion of the qualifier with
Car C, with the station-stop program disconnected, the car overshot the San Miguel
CL station (as expected) but then failed to shunt on the normally unused and
rusty rails . The car was stopped by the sand pile at the end of the track."
Had the implications of this incident been fully explored, the ."dead train"
detection problem might have been corrected earlier.

13 ECL 224

This incident also did not seem to have an effect on the redes i pi of
WELCO's system. Prior to the submission of bids to BART, wr.LCO redesigned
their train detection system employing the principle of control signal shunting.
We found no evidence that worst case analysis of train detection was studied.

The whole system of selection and testing prior to 1972 seemed deficient
concerning rail contamination and "dead trains". The system of the successful
bidder should have been tested prior to installation despite the time and cost
restrictions. This may seem to be unrealistic, yet it may effectively be
argued that safety considerations should not be limited by time and cost
restrictions if there is any doubt about the design.

A safety analysis report done by Battel le Memorial Institute for PBTB
was also deficient in this respect. This worst case analysis overlooked the
effect of loss of power to the train and rail contamination, as PBTB and WF.LCO
did. In fact, this report assumed detection and then proceeded to analyze
the remainder of the ATC system.

Finally, the role of the PUC must be included. As a public transportation
system, BART fell under the jurisdiction of the PUC for safety considerations.
It was the PUC's function to regulate BART to insure that the equipment design,
construction and operation be sufficiently safe. Since the BART system was
unprecedented in California, the PUC had no existing regulations for automatic
train control. The PUC therefore had to develop a set of regulations-
The result was General Order 127 of the Public Utilities Commission
of the State of California. This order was adopted on August IS, 1%7, and
became effective on September 15, 1967 .

Section 2 of General Order 127 stated that "no train protection equipment
or circuits in such equipment, which is part of an automatic train control

14 ECL 224

system, shall hereafter be constructed . . .until plans for such construction...
have been filed with and approved by the Commission. " Mien questioned on the
point of Commission responsibility, a BART District official stated that
General Order 127 did not go into effect until after the ATC contract was
awarded. It does seem reasonable to assume that WELCO started design of its
ATC before General Order 127 was established. The major part of the construction
was done after the date that the PUC had adopted General Order 127 and it seems
that the BART District did follow Article 2 of the general order after its
promulgation.

In 1972 when the Public Utilities Commission was conducting final tests

before letting BART commence operations, the failure of n dead train" detection

was realized. This failure to pass the "dead train" test meant that the

*

automatic train detection system would have to be. revised before operations
began because the worst case train protection had not been demonstrated to
the PUC.

To prevent further delay of starting operations and thus increased political
pressures, BART installed a manual block system. It was under this system that
BART was opened to the public in 1972. This manual block system required BART
personnel to be located at each station and in direct communication with the
other personnel at the respective stations. This system allowed for an
unsatisfactory 2 station headway. The procedure followed by station supervisors
was, that when a train left a station that was two stations ahead of the trailing
train, the supervisor would then call to the supervisor located at the station
of the trailing train and then, and only then, would the trailing train be
allowed to proceed.

15 ECL224

The manual block system was in service for two years before median lzi\X \ on
of the system occurred. Since the central computer at Lake Merritt was receiving
these occupancies and issuing door control commands, the fact that the
central computer could accomplish the same objective as the supervisors became a
reality. This system is called a Computer-Automated Block System (CABS). This
system utilized the fact that a train cannot proceed unless its doors are closed.
The central computer picks up the fact that the lead train (two stations ahead)
has closed its doors and has vacated the occupancy. At this time, the central
computer relays to the trailing train (two stations behind) that it may close
its doors and proceed. This exact system was known as CABS-2* Later the
system was refined so as to allow one station headways (CABS-1).

In order to achieve this new one station headway through the tmnsbay tube,
"ghost stations 11 were installed. M001 and M002, as these stations were called,
consisted merely of circuits in the tube where occupancy could be established
so as to allow the trailing train to enter the tube before the lead train had
exited, M001 and M002 were located on tracks headed in opposite directions.

An obvious constraint to this system is that headways cannot be closed up
to less than one station. This is true because the only place the central
computer can exercise control over the trains is at the stations. Another
disadvantage of the CABS-1 system was that if a train breaks down at some point
or for any reason is delayed, the trailing trains are still required by central
computer to maintain the one station headway, thus causing an eventual blockage
of the line. This one station headway is opposed to a 90 second headway,
dictated by the stopping distance required for the block.

In the short run, CABS-1 was an improvement over the manual block system
but BART still planned to rely on the primary train detection system in the
long run. As yet, the originally planned primary train detection is not

16 ECL224
controlling train separation. A system is now being planned which will use the
primary train detection system backed up by Sequential Occupancy Release (SOU).

SOR utilizes the principle that trains have to follow a logical, sequential
pattern. This is a memory-type system that has mini-computers around 1,500
locations (blocks) as compared to the 54 of the CABS-1 system. If a train
crosses a check point (mini-computer) for greater than one second, occupancy is
memorized by the computer and not released until occupancy is confirmed several
blocks ahead. This essentially is just an enlarged CABS system, but much more
effective because the computer actually creates occupancy. Precautions have been
taken for failure modes of the SOR as the system is more complex and computerized.
The computer might memorize occupancy in the wrong block or fail to reset an
occupancy after the train has vacated and cause a tie-up because of false
occupancy. This increased accuracy is not without cost. The additional hardware
and software increases the complexity and decreases the reliability.

BART officials are confident that the original train detection system, as
modified by SOR, can be utilized for train separation. This in turn will serve
to minimize headways. The primary systems along with SOR can achieve headways of
110 seconds, whereas the original primary system was to have brought headways
down to 90 seconds. This 20 second difference is the result of the computer
memorizing the occupancy and not resetting until the train is several blocks
down the line. Since these computers do memorize occupancy, they provide
additional safety during intermittent losses of detection.

In summary, the root of BART's ATC problems can be traced to unsatisfactory
engineering design and test procedures. No one party can he held totally
responsible. Rather, several involved agencies are responsible for completely
overlooking the problem of M dead train" detection, Westinghouse Electric Corporation
and Parsons-Brinckerhoff-Tudor-Bechtel, and the Public Utilities Commission of

17 ECL 224

the State of California all failed to consider the detection problem for a period
of almost seven years prior to BART's opening in 1972. This was a serious
oversight concerning the safety of the system and is reflective of poor
engineering. It was only brought to light during qualification testing by the
Public Utilities Commission of the State of California.

It is understood that in projects of the magnitude of BART there are
constraints that limit engineering design. Politics and economics sometimes
tend to oppose ideal safety conditions, yet all parties had a commitment to the
public safety. The State's regulating agency is immune to economic pressures
and it should insure the safe design and proper testing of the automatic train
control. The PUC, although late in exercising its authority, did properly
institute operating restrictions as outlined previously.

Arguments citing time and money constraints become meaningless when they
relate to solving problems of such primary concern as the safety of a large
number of passengers. Such problems must be resolved with subsequently larger
expenditures of time and money, The time and. monetary loss of supplementing
the original design with the CAB and SOR systems has not been the only effect
of this engineering inadequacy. More importantly, BART has suffered a loss of
support not only from the professional ranks but also from the tax paying public
and thus has negatively affected the movement toward a much needed large scale
rapid transit.

18

ECL Z24

Bibliography

1. San Francisco Bay Area Rapid Transit District .Standard Specifications -
Parts JSK, August 1966, Parsons-Brinckerhoff-Tudor-Bechtcl , San Francisco,
Calif.

2. San Francisco Bay Area Rapid Transit District Technical Reports - Technical
Reports on the BART District Test Track Program - 12 reports, 1966-67;
Technical Report 1: Automatic Train Control; Technical Report 12: Summary,
Parsons-Brinckerhoff-Tudor-Bechtel , San Francisco, Calif.

3. General Order No. 127 of the Public Utilities Commission of the State of
California, September 15, 1967.

4. Correspondence between Parsons- Brinckerhof f- Tudor- Bechtel and Mr. David G.
Hammond, Director of Development and Operations, BART District, dated
March 16, 1967.

5. Correspondence between Parsons- Brinckerhof f-Tudor- Bechtel and Mr. David G.
Hammond, dated April 11, 1967.

6. Correspondence between Parsons- Brinckerhoff -Tudor- Bechtel and Mr. B. R.
Stokes, General Manager, BART District, dated July 19, 1973.

7. IEEE Spectrum, September 1972, "The Grand Scheme", Gordon D. Friedlander. .

8. IEEE Spectrum, October 1972, "BART's Hardware - from Bolts to Computers",
Gordon D. Friedlander.

9. IEEE Spectrum, November 1972, "More BART Hardware", Gordon D. Friedlander.

10. IEEE Spectrum, March 1973, "Bigger Bugs in BART", Gordon D. Friedlander.

11. IEEE Spectrum, April 1973, "A Prescription for BART", Gordon D. Friedlander.

12. Electronics, July 26, 1965, "Who's on the Right Track", K. B. Riley and
L. S. Gomolak,

13. Aviation Keek and Space Technology, October 21, 1974, "Controversy Still
Clouds BART Program", Benjamin M. El son.

19

ACKNOWLEDGMENTS

ECL 224

We would like to extend our gratitude to the following men, without—
whose cooperation and information this case study could not have been
possible:

Krishna V. Ilari, Manager of System Engineering, Equipment Division,

Bay Area Rapid Transit (BART) Division;
Robert P. Townley, Train Control Engineer, Equipment Division,

Bay Area Rapid Transit (BART) Division;
Leo L.Lee, Senior Electrical Engineer of the Transportation Division

of the Public Utilities Commission.